CCE banner
Funded Research

Surface Weather Data with Uncertainty Quantification for Terrestrial Ecosystem Process Models

Thornton, Peter: Oak Ridge National Laboratory (Project Lead)
Vose, Russell: National Oceanic and Atmospheric Administration (Institution Lead)

Project Funding: 2013 - 2015

NRA: 2012 NASA: Terrestrial Ecology   

Funded by NASA

We propose to extend and improve an existing surface weather data product, providing essential land-atmosphere boundary conditions to terrestrial ecosystem process models. Our approach delivers a rigorous quantification of data product uncertainty that can be directly incorporated in modeling analyses. Our team recently released a first-ever surface weather kilometer spatial resolution and daily temporal resolution over the conterminous United States, Mexico, and southern Canada, for a 32-year period of record (1980-2011) ( This dataset, referred to as Daymet, was originally designed to provide forcing data for land process models and continues to serve that community, particularly in support of the demanding requirements for regional and continental-scale high-resolution ecosystem simulation. In the process of generating this data product, we have identified a number of targeted extensions and improvements that would significantly advance its utility in meeting the research goals of the terrestrial ecosystem modeling community. We have identified four areas where rapid progress can be made to address specific needs for surface weather products in the land process modeling community: 1) application of current Daymet algorithms to regions in Europe, Asia, and Australia supported by existing compilations of station data; 2) targeted augmentation of current surface observation synthesis products by merging multiple existing datasets under a consistent format; 3) modification of current Daymet algorithms to enable application in regions with low density of surface observations; and 4) expansion of existing Daymet algorithms to provide longwave radiation and sub-daily temporal resolution. The surface weather variables targeted in this proposal are precipitation (rain and snow), temperature, shortwave radiation, longwave radiation, and humidity. The current Daymet approach does not provide longwave radiation or surface winds. We propose to extend the approach to make estimations of longwave radiation, but estimations of high-resolution surface winds will remain beyond the scope of this effort. Detailed uncertainty quantification metrics are generated for the current Day met products, and we propose to extend these metrics to include new regions, and to take advantage of new independent observational data sources. Our fundamental uncertainty quantification method uses cross-validation to estimate spatial and temporal variation in estimation bias and prediction error for precipitation, temperature, humidity, and radiation. Data service, management, and usability have been constantly in focus for the Daymet effort from its outset, and we will continue this emphasis in the proposed work. Our team has forged strong ties with the NASA-sponsored Oak Ridge National Laboratory Distributed Active Archive Center (ORNL DAAC) for Biogeochemistry Dynamics. Our efforts will follow the best practices identified by the ORNL DAAC for data and metadata formats, dissemination and data sharing. Current Daymet products are currently indexed by the ORNL DAAC and are slated for integration with DAAC data delivery tools over the coming year. We intend to provide new Daymet data products emerging from the proposed activity as additional data resources indexed and potentially served by the DAAC.

2015 NASA Carbon Cycle & Ecosystems Joint Science Workshop Poster(s)

  • Gridded daily surface weather for North America: development and uncertainty analysis of the Daymet dataset   --   (Peter E. Thornton, Michele M. Thornton, Benjamin W. Mayer, Robert B. Cook, Ranjeet Devarakonda, Yaxing Wei, Suresh Kumar Santhana Vannan)   [abstract]

More details may be found in the following project profile(s):